[Bioc-devel] Problems with BiocParallel on Windows

[Jiefei Wang]

There is no need to do it in MulticoreParam as it will simply fork
your master process and therefore inherit your global setting. I think
the warning you get is due to a bug in RStudio, I hope they can fix it
but it might not be that fast...

Best,
Jiefei

On Mon, Oct 25, 2021 at 8:16 PM Giulia Pais <giuliapais1 using gmail.com> wrote:
>
> Sorry, my bad, I actually changed the code but for it work properly it is mandatory to rebuild the package completely (not only loading it through devtools). It seems to work for 1 function, I'll change the option export global in all the code and rebuild it and see if it works. Do you know if it is necessary also for MulticoreParam? It doesn't seem to be necessary and when I previously set the option to TRUE I always got warning messages for the stats package, that’s way we set it to FALSE.
> Thank you so much for your help
>
> On 10/25/21, 13:58, "Jiefei Wang" <szwjf08 using gmail.com> wrote:
>
>     I do not see it from the build report here:
>     http://bioconductor.org/checkResults/devel/bioc-LATEST/ISAnalytics/riesling1-checksrc.html
>
>     Can you please provide a reproducible example? It can be very helpful
>     to locate the issue
>
>     On Mon, Oct 25, 2021 at 7:53 PM Giulia Pais <giuliapais1 using gmail.com> wrote:
>     >
>     > I mean that trying to modify the code of one of the functions that raises errors and setting the option "exportglobals = TRUE" on SnowParam, somehow it still fails and says it doesn't find the function "mandatory_is_vars()" which is a function exported by the package
>     >
>     > On 10/25/21, 13:49, "Jiefei Wang" <szwjf08 using gmail.com> wrote:
>     >
>     >     What do you mean by "exported functions are invisible to SnowParam
>     >     workers"? Please provide more details. If you call 'bplapply' with a
>     >     function from your package, it should be visible to the worker.